Choke input filter LF noise

[piano3]

nicely put!!

[DF96]

Series resistance will damp any resonance, and improve the low pass filtering action especially at very low frequencies where the choke is giving up. Provided the resonance is reasonably well-damped, you then need to think about either more filtering or active solutions like regulators.

[Rundmaus]

The described filter chain in my case suffers from a slightly high Q-factor, but as long as I do not encounter problems with it, I am not willing to burn precious B+ voltage in additional series resistors...

[piano3]

Thanks DF, you have brought up several points I have not considered and with your usual clarity.

[DF96]

You can also damp the resonance with a snubber (C+R) to ground. No extra voltage drop that way. Effectively the snubber is an AC-only resistance. We usually think of snubbers as working at RF frequencies, but the principle applies to subsonic too. You just need much bigger caps!

[Rundmaus]

Can you go into more detail on that?

*interested*

[DF96]

I don't know the design equations for a snubber off the top of my head, but I'm sure Google will help. Essentially the resistor needs to be the right size to damp the resonance, roughly making up what is lacking from the load. Too big and it won't damp enough. Too small and the PSU will be slow to recover from demand peaks. Then the capacitor in series with it needs to be big enough to ensure that around the resonance frequency the resistor is 'seen' by the resonance. Too big and you almost have an extra smoother; too small and it won't do much damping.

You can either find the equations, or play with it in Spice. Or do both. That is about the extent of my knowledge - I would have to start doing algebra to get any further and I'm sure someone will have already done it somewhere.

[trobbins]

You may want to consider the topic from a few different angles.

Any low frequency noise getting through the power supply filtering will get attenuated through an OT, and also attenuated on its way to any preamp stage assuming there is resistive dropping to those stages. Assuming the magnitude of the <f(mains) noise is less than your mains ripple, then it's importance to amp performance is similarly proportionally less important.

If the low frequency noise is mainly in the form of disturbances, and not periodic, then each disturbance introduces a form of step disturbance to the LC filter and hence higher frequency components will be introduced. If the Q of the LC, or CLC, sections is >1 then any disturbance frequency component within the Q peak of the filter will be amplified in level, and depending on levels and Q could become significant at the output of the power supply.

Managing the Q of the filter can also be done using parallel R - which is a common technique for dampening higher frequency self-resonances within the choke - it does somewhat reduce the filter attenuation at the mains ripple frequencies. Even more exotic parallel R-zener(back-to-back) can be used for managing large signal disturbances, but I doubt you are experiencing low frequency disturbances with a magnitude across the choke that is much larger than mains ripple.

Imho, applying RC or resonant trap sections to the LC filter could be an onerous path to take, and possibly only applicable if you have a significant repetitive specific noise issue that can benefit from a specific tailoring of the LC filter response.

Ciao, Tim